Open Radar

 
Summary:
By using type inference and method overloading, CGFloat will be a lot easier to handle with 64bit compatibility.

Steps to Reproduce:
To compute the distance between two CGPoint instances, it is useful to use hypot()/hypotf() from <math.h>.

Right now, one has to write this code to make sure both 32bit and 64bit builds work well:
I want to calculate the distance between two CGPoint with hypot():
let distance : CGFloat = hypot(from.x - to.x, from.y - to.y)

Expected Results:
There are two overloaded global functions called “hypot()”:
func hypot (x: Double, y: Double) -> Double
func hypot (x: Float, y: Float) -> Float

Using type inference and overloaded methods like these, the difference between 64bit and 32bit architectures will blur and be a lot easier to manage.

Actual Results:
People have to create their own wrappers to make sure, they don't have to spread 32bit-vs-64-bit code everwhere in the code base.

#if arch(arm64)
let my_hypot : (CGFloat, CGFloat) -> CGFloat = hypot
#else
let my_hypot : (CGFloat, CGFloat) -> CGFloat = hypotf
#endif

Notes:
This also applies to several other functions such as sin/sinf, cos/cosf, etc.